Skip to content
Topos Uranos

Integration by change of variable

Summary

This lesson reverses the chain rule: if FF is an antiderivative of ff on an interval JJ and gg is differentiable on an interval II, with values in JJ, the composition F(g(x))F(g(x)) is an antiderivative of f(g(x)) g′(x)f(g(x)) \, g'(x) on II. That theorem, which is proved by differentiating, justifies the notation u=g(x)u = g(x), du=g′(x) dx\mathrm{d}u = g'(x) \, \mathrm{d}x, with which the computation looks like a substitution of symbols without being one. Its first consequences are the linear change of variable, which extends the table of basic antiderivatives to formulas with a parameter, and three forms worth recognizing at a glance: the quotient g′(x)g(x)\frac{g'(x)}{g(x)}, the power (g(x))rg′(x)\left(g(x)\right)^r g'(x) and the exponential eg(x)g′(x)e^{g(x)} g'(x); the first one yields the antiderivatives of the tangent and of the cotangent. The lesson also proves the inverse substitution, x=h(t)x = h(t), which requires a derivative without zeros in order to return to the original variable, and closes with criteria for choosing the substitution.

Learning objectives

  1. Prove the change of variable theorem with the chain rule, and use the notation u=g(x)u = g(x), du=g′(x) dx\mathrm{d}u = g'(x) \, \mathrm{d}x as an abbreviation that the theorem justifies.
  2. Derive the linear change of variable and the antiderivatives with a parameter, and recognize the forms g′(x)g(x)\frac{g'(x)}{g(x)}, (g(x))rg′(x)\left(g(x)\right)^r g'(x) and eg(x)g′(x)e^{g(x)} g'(x), with the interval on which each antiderivative holds.
  3. Prove the inverse substitution, explain why it requires a derivative without zeros, return to the original variable and extend the antiderivative to an endpoint of the interval when appropriate.
  4. Choose a substitution, rewriting the integrand if necessary, and compare antiderivatives of different appearance obtained by different routes.

The chain rule read backward

The table of basic antiderivatives from the lesson on antiderivatives and the indefinite integral reads the derivatives of the elementary functions backward, and linearity reads the derivative of a linear combination backward. Neither of them says what to do with ∫2xcos⁡(x2) dx\int 2x \cos\left(x^2\right) \, \mathrm{d}x. But that integrand is the derivative of a composition: by the chain rule, the derivative of sin⁡(x2)\sin\left(x^2\right) is cos⁡(x2)⋅2x\cos\left(x^2\right) \cdot 2x. The factor 2x2x is the derivative of the inner function, and the cosine, evaluated at the inner function, is the derivative of the outer one. To recognize such an antiderivative, one has to see in the integrand a function evaluated at g(x)g(x), multiplied by g′(x)g'(x).

TheoremIntegration by change of variable

Let JJ be an interval with more than one point, FF an antiderivative of ff on JJ, II an interval with more than one point and g ⁣:I→Rg \colon I \to \mathbb{R} a function differentiable at every point of II, with g(I)⊆Jg(I) \subseteq J. Then F∘gF \circ g is an antiderivative of (f∘g) g′(f \circ g) \, g' on II:

∫f(g(x)) g′(x) dx=F(g(x))+C,x∈I\int f(g(x)) \, g'(x) \, \mathrm{d}x = F(g(x)) + C, \qquad x \in I

One differentiates F(g(x))F(g(x)) with the chain rule. The figure shows, with f(u)=cos⁡uf(u) = \cos u, F(u)=sin⁡uF(u) = \sin u and g(x)=x2g(x) = x^2, the integrand 2xcos⁡(x2)2x \cos\left(x^2\right) and the antiderivative sin⁡(x2)\sin\left(x^2\right), with a tangent line whose slope is the value of the integrand, and then the vertical translations of that antiderivative.

Proof

  1. (F∘g)′(a)=F′(g(a)) g′(a)=f(g(a)) g′(a)(F \circ g)'(a) = F'(g(a)) \, g'(a) = \resaltar{f(g(a)) \, g'(a)}

    Let aa be a point of II. Since II is an interval with more than one point, aa is an accumulation point of II. The function gg is differentiable at aa, its values lie in JJ, and FF is differentiable at g(a)g(a), with derivative f(g(a))f(g(a)), by the definition of antiderivative (if g(a)g(a) is an endpoint of JJ that belongs to JJ, that derivative is the one-sided one, which is the derivative of FF as a function defined on JJ). By the chain rule theorem, from the lesson on the chain rule:

  2. ∫f(g(x)) g′(x) dx=F(g(x))+C,x∈I\int f(g(x)) \, g'(x) \, \mathrm{d}x = \resaltar{F(g(x)) + C}, \qquad x \in I

    The equality holds at every point of II: F∘gF \circ g is an antiderivative of (f∘g) g′(f \circ g) \, g' on II. By the theorem on the antiderivatives of a function on an interval, from the lesson on antiderivatives, the others differ from it by a constant.

The theorem does not require ff to be continuous, nor gg to be injective or to have a derivative without zeros: it requires ff to have an antiderivative on an interval containing the values of gg, and gg to be differentiable. In the example, g(x)=x2g(x) = x^2 is differentiable on R\mathbb{R}, its image [0,∞)\ico{0}{\infty} is contained in J=RJ = \mathbb{R}, where sin⁡u\sin u is an antiderivative of cos⁡u\cos u, and on R\mathbb{R}:

∫2xcos⁡(x2) dx=sin⁡(x2)+C\int 2x \cos\left(x^2\right) \, \mathrm{d}x = \sin\left(x^2\right) + C

The hypothesis g(I)⊆Jg(I) \subseteq J is what fixes the interval of the result: if FF is an antiderivative of ff only on JJ, the composition F∘gF \circ g is an antiderivative only where gg takes its values in JJ. What the theorem does not settle is recognizing, in a given integrand, the form f(g(x)) g′(x)f(g(x)) \, g'(x); the rest of the lesson is devoted to that.